University of Wisconsin-Madison, Department of Agricultural Economics Staff Paper Series. January 1994 No. 368

Transcription

1 Copyright (c) 1994 by Lisa C Smith. All rights reserved. Readers may make verbatim copies of this document for non-commercial purposes by any means, provided that this copyright notice appears on all such copies. University of Wisconsin-Madison, Department of Agricultural Economics Staff Paper Series January 1994 No. 368 STRUCTURAL ADJUSTMENT AND WELFARE IN RURAL AFRICA: THE ROLE OF RESOURCE CONTROL IN HOUSEHOLDS By Lisa C. Smith University of Wisconsin-Madison

2 OUTLINE I. Introduction II. Structural Adjustment in Sub-Saharan Africa III. Conceptual Framework: Welfare Provisioning in Households A. Model of Time and Income Allocation B. Gender-Differentiated Allocational Priorities, Access and Power C. Allocative Inefficiency IV. Structural Adjustment Policy Impacts on Household Welfare A. Expenditure Switching Policies B. Reductions in Government Expenditures on Social Services V. Conclusion

3 I. INTRODUCTION Structural adjustment programs supported by the International Monetary Fund (IMF) and the World Bank have been implemented in over thirty Sub-Saharan African countries. The goal of these programs is to rectify macroeconomic imbalances (fiscal and balance of payments deficits) and low economic growth associated with the economic crisis of the 1970 s. Until recently, the programs success was mainly gauged in terms of macro-level indicators, such as deficit reductions and increases in economic growth. For the effects of the policies on the people living in countries undergoing adjustment, it was assumed that economy-wide improvements would eventually lead to improvements in their welfare as well (Anyaoku 1989; Jolly 1988). By the early 1980 s it became apparent that the changes induced during structural adjustment can entail burdens for some groups of people and benefits for others: some people find themselves coping with the negative impacts of policy reforms, while others thrive by taking advantage of new opportunities for raising their incomes (Haddad 1991; World Bank 1991). Concern arose about the effects of the programs on economically poor groups of people in particular. In the 1990 s, stemming the tide of poverty and enhancing the welfare of the poor through promoting their active participation in bringing about economic growth and through improving their access to social services has become a fundamental objective of (the World Bank s) structural adjustment programs (World Bank 1993a). Numerous attempts have been made to gather evidence of the impacts of policies associated with structural adjustment programs on peoples welfare. Many studies give evidence of strong negative impacts of the policies. A study undertaken by the Commonwealth Secretariat (1989:70), for example, claims that structural adjustment policies leading to "reductions in household incomes, increases in food prices and cut backs in health services... have been the main causes for the deterioration in children s nutrition and health during the 1980 s observed in many countries." The study cites evidence of increased incidences of malnutrition during the early 1980 s in as many as twenty-five adjusting countries. Another study claims that there has been "a widespread deterioration in the nutritional status of children and pregnant and lactating mothers in both rural and urban areas in countries with IMF Stabilization and World Bank structural adjustment programs" (Elson 1989:16). 1

4 It is difficult, however, to verify whether current negative welfare patterns found in many adjusting countries result directly from structural adjustment policies themselves. They could stem from economic conditions and government policies that existed prior to the implementation of adjustment measures (Elson 1991; Behrman 1988). For this reason, in addressing welfare concerns surrounding the policies, economists in particular have taken an inferential as opposed to empirical approach. Detailed models linking macro-level changes to micro-level outcomes such as household incomes and consumption have been developed (see for example Sarris 1992; Sahn 1991). Such models have been helpful for identifying the potential impacts of structural adjustment policies on household-level variables. However, they have been criticized for their failure to treat sub-household level processes and non-market processes as important and significant determinants of welfare outcomes (Elson 1991; Csete 1992). In particular, they fail to consider differences in the constraints and decision making processes of individual household members, which can place limits on the ability of households as a whole to react to adjustment policies in a way that maintains and/or enhances household welfare. This paper contributes to the conceptual understanding of how structural adjustment policies impact rural Sub-Saharan African peoples welfare by exploring the policies intermediate impacts on the allocation of economic resources within households. Although the dependent variable of the study is household welfare, the study s examination of resource allocation is focused at the intrahousehold level. In particular, the differential resource allocation behaviors of women and men in households, including how they use the economic resources (time and income) they control to contribute to their households welfare and how they obtain these resources, are examined. Recognizing the existence of households in which there is only one main adult decision maker (the proportion of female headed households in rural Sub-Saharan Africa, for instance, is estimated to be 25 percent (Due 1991)), the paper concentrates on households containing both male and female adults as primary decision makers. 2

5 Throughout the paper, "welfare" is defined to be the physical well-being of human beings. 1 Variables most closely associated with physical well-being are peoples states of health and nutrition. Child and adult mortality rates in Sub-Saharan Africa, associated with low levels of health and nutrition, remain the highest in the world. In 1990, the number of lost disability-adjusted life years (DALYs), a measure combining the loss of life from premature death with the loss of healthy life from disability, was higher in Sub-Saharan Africa than in any other area of the world (World Bank 1993b). Attention to improving human welfare in Sub-Saharan Africa is timely and crucial, not only as a goal in itself, but also for maintaining the productivity of Africa s "human capital" in the coming decades. The effects of two types of structural adjustment policies on human welfare will be investigated in this paper. These are policies that promote the production of tradeable goods in agricultural sectors (expenditure switching policies) and policies that reduce public provision of social services (expenditure reduction policies). Two main questions are examined: (1) How are time and income allocated in households? A general conceptual framework for understanding the micro-foundations of welfare provisioning in rural Sub-Saharan African households is developed. After concluding that many intrahousehold level processes related to gender differences are fundamental to welfare provisioning in households, this conceptual framework is used to ask: (2) Given the intrahousehold level processes identified, what can be inferred about the possible welfare impacts of structural adjustment policies? To be considered are the avenues through which adjustment policies, mediated by their effects on resource allocation in households, impact peoples welfare. II. STRUCTURAL ADJUSTMENT POLICY IN SUB-SAHARAN AFRICA In the 1980 s many Sub-Saharan African countries experienced severe economic distress in the form of balance of payments and budget deficits, high inflation rates, and low economic growth. Although the causes of these problems have varied by country, four major contributing factors have been identified. The first is external shocks stemming from higher oil prices since the 1970 s, lower prices for countries primary commodities, and increases in the real interest rate on commercial debt. The second is environmental shocks resulting from 1 The paper thus departs from the use of the term welfare in the field of welfare economics in which, under the nonpaternalism postulate, a person s welfare is identified with the person s (own perception of) the utility they receive from any resource allocation (Boadway and Bruce 1984). 3

6 increased variability in rainfall patterns. The last two factors are over-expansionary fiscal and monetary policies and domestic pricing policies biased against agriculture, the sector which is the greatest source of production in most Sub-Saharan African countries. Structural adjustment 2 policies are aimed at rectifying macroeconomic imbalances and restoring sustainable economic growth. They include relatively short-term stabilization policies, which have the specific goal of reducing inviable fiscal and balance of payments deficits. They also include longer term policies that address the more fundamental distortions underlying macroeconomic imbalances, such as policies to improve resource allocation, increase economic efficiency, expand growth potential and increase resilience to shocks. The three main types of structural adjustment policies implemented in Sub-Saharan African countries are currency devaluation, liberalization of product markets and reductions in government expenditures. Government expenditure reduction is aimed at reducing government budget deficits. It may take place in the form of reductions in expenditures on social services (for example, health and education services and staple foods subsidization) and/or on economic infrastructure (for example, road building and repair and investments in marketing facilities). Devaluation and liberalization policies, together referred to as "expenditure switching" policies, induce a fundamental restructuring of a country s product and factor markets. The goal of currency devaluation is to depreciate an economy s real exchange rate (the price of nontradeables relative to tradeables 3 ) inducing a shift from the production of nontradeable products into the marketed production of tradeable products. Devaluation also removes rationing in the foreign exchange market and increases supplies of consumer goods, giving producers incentives to increase their output for sale on the market. The goal of liberalization is to bring the domestic relative prices of tradeable products into line with world relative prices through removing controls on trade, such as import controls, tariffs, or export taxes and subsidies. Removing such controls generally leads to 2 The term "structural adjustment" is used to encompass both stabilization and structural adjustment policies. 3 If the price of a product is determined by world market conditions, it is classified as a "tradeable" good. Tradeable products produced domestically include exports and unprotected import substitutes. If the price of a product is determined by domestic supply and demand, the product is classified as "nontradeable". 4

7 an increase in a country s terms of trade (the domestic price of exports relative to imports), shifting output towards the production of exports and unprotected import substitutes, which are tradeables, and out of protected imports substitutes, which are nontradeables (World Bank 1991). Thus, both devaluation and liberalization are expected to induce increases in the production of tradeable relative to nontradeable products. In order for such a shift to come about, reallocations of labor and other factors of production from nontradeable to tradeable production must take place. Agriculture is the most important tradeable sector in Sub-Saharan African countries. It is in the agricultural sector that most export production and tradeable food production takes place. At the same time as it is the site of the greatest potential increase in tradeable production, the agricultural sector is the source of employment, income and consumption for the large majority of people in Sub-Saharan Africa. In the following section a conceptual framework for tracing the impacts of expenditure switching and expenditure reduction (specifically, reductions in expenditures on social services) policies on household welfare--including children s, men s and women s welfare--will be developed. 5

8 III. CONCEPTUAL FRAMEWORK: WELFARE PROVISIONING IN HOUSEHOLDS Most agricultural production in Sub-Saharan Africa takes place on family farms, in which relations between people who are linked through family ties, including marriage, are important for the allocation of labor and other inputs into production. The African household is commonly composed of several nuclear families, each made up of a husband, one or more wives, and the children born to each wife. Complex economic interactions between members of separate nuclear families and, within families, between wives of the same nuclear family, may take place. In the following conceptual framework, this complex reality is sacrificed in order most clearly to illustrate the significance of the distinct behaviors of women and men for welfare provisioning in family-based production units of rural Sub-Saharan Africa. I consider a two decision maker economic unit or "household," made up of a woman and a man (the decision makers) and their children. I also assume that production takes place with only the use of the decision makers labor. The conceptual framework is presented in diagrammatic form in Figure 1, which will be referred to throughout this section. Figure 1 shows, for our hypothetical two-person household, links between time allocation, income generation, income allocation, and household welfare. A. Model of Time and Income Allocation Patterns of time and income allocation in households are thought to be founded on social roles underlying household members obligations to provide welfare. Social roles that are germane to welfare provisioning can be usefully classified into the categories of production and reproduction. The role of production is associated with the transformation of physical product to be used for direct consumption by household members, for exchange, or for sale on the market. The role of reproduction, which is directly associated with peoples physical well-being, can be further broken down into subcategories of generational reproduction and daily reproduction. Generational reproduction involves the bearing and early nurturing of infants and the care and upbringing of children. Daily reproduction involves recurrent tasks of securing the dayto-day 6

9 7

10 material survival of the household, such as cooking, collecting water and firewood, and cleaning (Ellis 1992). These role categories will be employed below to provide useful distinctions between categories of time and income allocation. Time Allocation Time is the most basic resource with which welfare is provided in households. How household members time is allocated among different activities is thus key to human welfare. Individuals may spend time in the provision of services to others and to themselves by engaging in activities that directly enhance their and other household members physical well-being, i.e., in "reproductive" activities. Time thus spent falls under the role category of reproduction. Individuals may also provide welfare indirectly by contributing to the availability of what will be termed "welfare inputs," defined to be material goods that enhance the physical wellbeing of household members. Time thus spent falls under the role category of production. Examples of welfare inputs are food, clothing, and medicine. Welfare inputs can be provided in two ways. The first is through time spent producing goods, such as food, that are directly consumed by household members. Such time is referred to as time in "production for consumption". The second is through time spent in production for cash income, the income from which can be used to purchase welfare inputs. Reproductive activities and production for consumption are classified as "nonmarket" work, production for cash income as "market" work. For our two-person hypothetical household, Figure 1 shows the links between, on the one hand, household members time allocated among reproductive activities, production for consumption and production for cash income (market work) and, on the other, household welfare. Time spent in reproductive activities feeds directly into household welfare. Time spent in production for consumption feeds indirectly into household welfare through contributing to the quantities of goods available for consumption by household members. Time spent in market work also ultimately feeds indirectly into household welfare through contributing to goods consumed by household members. However, the manner in which the cash income generated is allocated among purchased welfare inputs and other goods depends on a complex decision making process involving transfers of income between household members. This process will be elaborated below. 8

11 The allocation of household members time among the activity categories delineated above and their non-work time is constrained by their time endowments (twenty-four hours a day). Denoting the woman "agent f" (indexed i=f) and the man "agent m" (indexed i=m), household members time constraints can be characterized as: [1] where T i R = agent i s time in reproductive activities = agent i s time in production for consumption i T M = agent i s time in market work i T o = agent i s non-work time T = agent i s time endowment. T PC i Income Generation and Allocation In rural Sub-Saharan African households, agricultural production commonly takes place in a combination of joint, or "communal," and individual, or "personal," production activities. In personal production the majority of labor is provided by one individual household member; in communal production, labor is provided by all active household members. 4 Traditionally almost all of both communal and personal agricultural production was of food for direct consumption by household members. In modern times in most areas of Sub-Saharan Africa, at least some agricultural produce is sold on the market for cash income. In addition, many people have become involved in non-agricultural income generating activities such as food processing, handicraft production for sale, and trade (these are mainly personal production activities). A communal-personal dichotomy in labor allocation to income generating activities, whether they be agricultural or non-agricultural, has evolved parallel to subsistence agricultural production. 4 This common pattern is described in Whitehead (1990), Saul (1989), McMillan (1986), Roberts (1988), Koopman (1991), Guyer (1981), Davison (1988), David (1991), Webb (1989), and Dey (1992). Although in some areas men and women exchange labor in their personal production activities, in this analysis such labor exchanges will be ignored. 9

12 For our two-member household, agents time in market work (T M i ) can thus be further disaggregated into personal (MP) and communal (MC) market work (see Figure 1): [2] Composite production functions for communal and personal income generating production, respectively, can be characterized as [3] [4] where V MC and V i MP are purchased inputs into communal and personal income generating production. In Sub-Saharan African households, cash incomes received by different household members from income generating production are generally not pooled (Guyer 1988; Fapohunda 1988). Social norms govern who in households receives income from different activities and who pays for productive inputs. In general, for communal production this individual is a man, while for personal production it is the individual earner, which may be a man or a woman (see references in footnote 4). Although incomes received by household members are not generally pooled, household members often make transfers of cash income to each other. 5 Thus, their income constraints can be characterized as being separate, yet linked through income transfers. Let the categories into which cash income is allocated by agents m and f be their purchases of welfare inputs (denoted X W i ), their purchases of inputs into income generating production (V MP f for agent f; V MP m and V MC for agent m), and their purchases of "other goods" (denoted X o i ). 6 Agents income constraints restrict their total expenditures on these goods to 1) the income they receive from income generating production, plus 2) the income transfers they receive from outside of the household (their exogenous incomes), plus 3) the income 5 Discussion of income transfers can be found in Guyer (1988 and 1989), Jones (1986), Afonja (1990), David (1991), Roberts (1988), and Whitehead (1990). 6 These variables represent composite commodities combining all of the agents purchases falling into each category. 10

13 transfers they receive from the other agent. Let t be the net cash income received by agent f from agent m. Then agent f s and agent m s income constraints, respectively, can be characterized as [5] [6] where E i = agent i s exogenous income p i w = the price of welfare inputs purchased by agent i i p o = the price of other goods purchased by agent i v i = the price of purchased inputs into agent i s personal income generating production q i = the price of output from agent i s personal income generating production v c = the price of purchased inputs into communal income generating production q c = the price of output from communal income generating production Resource Control and Decision Making Resource control can be defined as the ability to decide how a resource is allocated. In the case of time allocation control is the ability to decide how time is allocated among different activities. In the case of income allocation control is the ability to decide how income is allocated among different purchases. Thus, resource control in households is germane to the decision making process in them. The degree of control over time and income by different household members is influenced by social norms governing decision making and by the balance of bargaining power in households. In the literature on decision making in Sub-Saharan African households two major themes emerge. The first is that men and women often function in separate economic spheres in which they make independent decisions regarding the allocation of resources they control (Whitehead 1990; Guyer 1986; Saul 1989; Fapohunda 1988; Palmer 1988). The second is that with the rise of commercialization has come increased conflict between women and men over the activities and purchases to which time and income are allocated. This conflict is manifested in negotiation between women and men over resource control itself. As noted above, the product of agricultural production traditionally remained primarily in the form of food. Customs and norms governed the distribution of this product to meet the consumption needs of households and their social obligations. With the rise of commercialization has come sales of agricultural produce and new non-agricultural income generating activities, along with a range of goods for purchase on the market. 11

14 Consequently, a portion of households product is in the form of cash income rather than food, and a selection of goods to purchase with this income is available. Money, as a generalized means of exchange, typically brings with it wider choices for allocation than does physical product. The transformation from subsistence to semicommercialized households has brought with it conflicts among household members over who earns income, who controls it, and how it is allocated. 7 Thus, although men and women commonly function in separate spheres of economic activity in which they make independent decisions, at times they are brought into a joint decision making unit in which conflicts are resolved. Social norms governing decision making over the allocation of inputs into communal and personal production influence control over labor in households. The primary decision maker for communal income generating production is usually a male adult, who makes decisions regarding the allocation of household members labor and other productive inputs. For personal income generating production the individual earner is the primary decision maker for allocation of labor and other inputs. However, because men receive the income generated in communal production while women contribute labor to it, the conflict between women and men surrounding income is manifested in negotiation between them over the time women spend in communal income generating production (T f MC ) and over transfers of income between women and men (t). Because neither men nor women make completely independent decisions regarding the allocation of women s labor to communal income generating production (this is negotiated), neither have full control over it. 8 Although women do not fully control the amount of time they spend in communal income generating production, they do make independent decisions over the allocation of their time in other activities. In terms of agent f s time constraint (equation [1] for i=f), the time agent f controls independently is the left hand side of 7 Discussion of such conflicts and some examples are given in Whitehead 1990, David 1991, Dey 1992, Guyer 1989, Holmboe-Wattesen and Wandel 1988, Jones 1986, Obbo 1990, Roberts 1988, Saul 1989, and Pankhurst See references in footnote 4 and Guyer 1981, Roberts 1988, and Koopman 1991 for viewpoints on men s control of women s labor in Sub-Saharan African households. 12

15 [7] Since men generally make independent decisions regarding the allocation of their time among all activities, they fully control their own time (the left hand side of equation [1] for i=m). Social norms governing who receives income from different income generating activities have been described above. Income accrual of household members is translated into income controlled by them through the mediation of income transfers. In general, a person who receives income from any activity is likely to be the person who controls it. However, where income transfers take place and are negotiated over, both men and women do not independently control all of the income they receive. By definition, they do make independent decisions over the allocation of the income they control, i.e., the income they receive net of transfers (the right hand sides of equations [5] and [6]). The outcome of negotiation between agents with differing interests is thought to be determined by their relative bargaining powers. Bargaining power gives a person the ability to influence the outcome of negotiation with others in the person s own interests. In general, in the household context, bargaining power is thought to be determined by the opportunity cost of being a member of the household, or how "well" individual members could do if they were to leave the household. A significant economic determinant of this opportunity cost is thought to be the ability to earn income independently of other household members. Such ability gives a person a measure of independence from other household members, increasing their capability of providing for themselves if they leave the household (Folbre 1988; Pollack 1985). Figure 1 lays out the process through which time spent by men and women in market work feeds into household welfare. The dashed lines indicate an arena of negotiation. Both women and men provide labor to communal market work. With changes in households economic environments such as those which come with commercialization, negotiation takes place over women s labor allocated to communal market work. Men and women also provide labor to personal market work. Men receive income earned in communal market work and in their own personal market work, while women receive income earned in their own personal market work. The income controlled by women and men is made up of the income they receive from market work net of income transfers between them, over which negotiation takes place. Finally, men and women make independent 13

16 decisions over the allocation of income they control to purchased goods. If these goods are welfare inputs, they enhance household welfare. The decision making process described above involves a combination of unilateral and joint decision making. Joint decision making (negotiation) takes place over the time and income controlled by different household members, i.e., over the distribution of resource control, while unilateral decision making takes place over the allocation of time and income they control. In the following section a formal mathematical model of this decision making process is presented. Model of Decision Making Let welfare provisioning in households be governed by the following function: where [8] [9] 14

17 The inputs into welfare provisioning are women s and men s purchased welfare inputs (X i W ) and the time they spend providing welfare, either directly through reproductive activities (T i R ) or indirectly through production for consumption (T i PC ). 9 Let agents allocational priorities among household welfare (W), other purchased goods (X i o ) and their and the other agent s non-work time (T i o ) be governed by utility functions: [10] In the model presented below, a game theoretic approach is taken, which allows modeling the decision making of multiple agents with differing interests. Game theoretic models can be classified into two types, noncooperative and cooperative. In noncooperative models each agent makes decisions unilaterally (given their expectations of other agents choices), while in cooperative models agents make decisions jointly. 10 The complexity of a decision making process involving both unilateral and joint decision making (the latter determining constraints for the former) can be simplified for modeling purposes by separating them into a twostage game theoretic framework as do Carter and Katz (1994) in their "Conjugal Contract" model. In the first stage, choice over the unilaterally-decided variables takes place. This is modeled as a noncooperative game. 11 Agents choose their allocation of time to welfare provisioning (T W i ), personal market work (T MP i ), and non-work (T o i ) unilaterally. They also choose the allocation of the income they control among welfare inputs (X W i ), productive inputs (V MP i ), and other goods (X o i ) unilaterally. In addition, agent m 9 Household welfare is treated as a "z-good" as in Household Production Models (Becker 1965). Time in welfare provisioning could be modeled as entering indirectly into equation [8] as an input into a large number of production functions for produced welfare inputs. These could be, for example, food, cooked food, clean water, or child care. To simplify, the exact products being produced (in the case of T PC i )or service being provided (in the case of T R i ) are not given explicit production functions. It is assumed that household members take into account the myriad of goods and services into which their welfare provisioning time enters as they allocate their time to welfare provisioning (T W i ). 10 Cooperative bargaining models were first adapted to the analysis of household decision making by Manser and Brown (1980) and McElroy and Horney (1981). Jones (1983) used a cooperative bargaining model for analysis of intrahousehold resource allocation in the African context. Noncooperative game theory has been used for the analysis of household decision making by, among others, Chiappori (1992), Ulph (1988), and Lundberg and Pollack (1992). 11 In particular, it is modelled as a two person strictly competitive game of complete information, meaning that no coordination between agents is needed for equilibrium strategies to be chosen (Carter and Katz 1994). 15

18 unilaterally chooses agent m s time allocated to communal market work (T m MC ) and income allocated to purchased inputs into communal income generating production (V MC ). In the second stage, choice over the jointly-decided variables takes place. These are the net transfers of income from agent m to agent f (t) and the amount of time agent f spends in communal market work (T f MC ). Choice over the jointly-decided variables is modeled as a Nash cooperative bargaining game in which agents negotiate over alternative Pareto optimal outcomes by maximizing a Nash objective function. Mathematically the objective function is the product of the agents gains from membership in the household, represented by the differences between their utilities in the current state and their maximum possible utilities if the household unit were to dissolve. The latter are referred to as "fall-back positions". The Nash objective function reflects the bargaining power effects of the opportunity cost of being a household member: the more attractive an individual s opportunities outside of the household, the more strongly are the individual s allocational priorities reflected in the final levels of the jointly-decided variables. Fall-back positions can be represented mathematically by specifying an indirect utility function for each household member giving maximum utility levels if they are no longer part of the household unit. 12 For women, economic variables affecting these fall-back positions are the prices associated with their personal income generating activities, the prices of the products they would purchase if they were to leave the household, and their exogenous incomes. For men, along with these variables are added the prices associated with communal income generating activities. The fall-back positions are determined by maximization of agents utility functions subject to the time and income constraints they would face if they were to leave the household. 12 Divorce is an option in many areas of Sub-Saharan Africa. See Pool 1972, Trincaz 1983, Guyer 1990, David 1991; Hanger and Moris 1973 and Capron 1973 for discussion of and examples of marital instability in African families. 16

19 Let the fall-back positions for agent f and agent m, respectively, be: [11] [12] where p i, i=f,m are the prices of goods purchased by the agents if they are no longer part of the household unit. The model is as follows. STAGE ONE: Unilateral Decision Making Over Allocation Each agent s utility function is maximized given fixed levels of (or given their expectations of) the other agent s choice variables and the jointly-decided variables. Agent f chooses T f W,T f o,t f MP,X f W,X f o, and V f MP to maximize subject to Agent m chooses T W m,t o m,t MP m,t MC m,x W m,x o m,v MP m and V MC to maximize subject to where the productions functions are given in equations [3] and [4]. This process yields a system of reactions functions, each giving agents choice variables as functions of the other agent s choice variables (along with the jointly-decided variables). Solved simultaneously, the system 17

20 yields a reduced-form equation for each unilaterally-decided choice variable conditional on the levels of the jointly-decided variables (which are given as parametric in the first stage). The conditional reduced-form equations are [13] [14] [15] where [16] These equations give the optimal levels of the unilaterally-decided variables for each possible (t,t f MC ) combination. STAGE TWO: Joint Decision Making Over Distribution Agents jointly choose t and T f MC to maximize subject to [17] [18] where the fall-back positions are given in equations [11] and [12], and equation [17] is the Nash objective function. Equation [18] limits resource allocation to outcomes in which both agents have positive utility gains from membership in the household. Equilibrium reduced-form equations for the jointly-decided variables are: where 18

21 [19] [20] The optimal levels of income transfers and of agent f s time in communal market work are "constrained Pareto optimal." That is, given that time and income controlled by agents are optimally allocated to unilaterallydecided variables (from the point of view of their allocational priorities), no agent can be made better off without the other agent being made worse off. Reduced-Form Equation for Household Welfare Substituting equations [19] and [20] into equations [13] in turn yields equilibrium reduced-form equations for time and income allocated to welfare provisioning: [21] [22] Finally, substituting equations [21] and [22] into equation [8] yields an equilibrium reduced-form equation for household welfare: [23] Without going into the optimality conditions and comparative statics effects which the above model generates, 13 immediately apparent from equation [23] is the separate, rather than additive, appearance of the exogenous components of agents incomes. Exogenous incomes enter separately into equation [23] for two reasons. First, agents do not pool their incomes. They make decisions over their expenditures independently 13 A more formal analysis of similar models can be found in Katz (1992), Carter and Katz (1994), and Smith (forthcoming). 19

22 subject to separate budget constraints. Second, the incomes that accrue to them, including their exogenous incomes, enhance their bargaining powers. Equation [23] is derived from a complex mathematical model of decision making, which remains a gross oversimplification of reality. Its usefulness lies in permitting identification of key contact points between changes in households economic environments and household welfare. According to the model, changes in the prices (and exogenous incomes) faced by household members affect household welfare through changes in both market and non-market time allocation and, mediated by changes in income controlled by household members, through changes in household members expenditures on purchased goods. B. Gender-Differentiated Allocational Priorities, Access and Power In the model presented above, agents are treated symmetrically with the exceptions that 1) agent m s allocation of labor to communal production is a unilateral decision while the allocation of agent f s labor to it is a joint decision and 2) agent m receives the income from communal production and purchases inputs into it. Three social factors underlay additional differences in the patterns of men s and women s time and income allocation in rural Sub-Saharan African households. The first factor is women s and men s differing responsibilities for allocating time to welfare provisioning (T i W, i=f,m) and for allocating income to welfare inputs (X i W, i=f,m). These responsibilities influence their allocational priorities. The second factor is men s and women s differential independent access to productive resources. The last factor is differences in women s and men s bargaining powers. Men and women in Sub-Saharan African households tend to have distinct individual responsibilities for family maintenance (Guyer 1988). Many differences in the responsibilities accorded to them are thought to be founded on their differing social roles. Biophysical circumstances account for women s almost exclusive role in generational reproduction: only women can bear and breast feed children. Although both women and men are physically capable of carrying out tasks associated with the role of daily reproduction, in most societies this role is joined with generational reproduction as the almost exclusive domain of women. The role of production is by contrast shared by both men and women (Ellis 1992; Joekes et al 1988). 20

23 The role of reproduction is more closely linked with peoples physical well-being than is the role of production. Correspondingly, women tend to be more directly involved in welfare provisioning than men (Lele 1991; Commonwealth Secretariat, 1989). Evidence from time allocation studies reveal that a far greater amount of women s time is spent in reproductive activities than men s (i.e., T f R >T m R ). By contrast, both women and men are highly engaged in productive activities, i.e., in production for household consumption and/or in market work (World Bank 1991; Davison 1988), with differences in their time spent in these activities varying by region. Given such a division of labor in households, male and female labor can be characterized as having low substitutability in reproductive activities (Ellis 1992). With regard to income allocation, the general consensus is that (in developing countries) men tend to use a larger proportion of income than women for productive, household maintenance and social investment, and for their personal consumption; women tend to use a larger proportion of income than men for meeting daily household consumption needs, such as food, clothing and health care (Kennedy and Bouis 1993; Joekes et al 1988; Blumberg 1989; Bruce and Dwyer 1988; Lele 1986; Koopman 1991). In many areas of rural Sub-Saharan Africa women are responsible for paying for food while men are responsible for paying for clothing, medical expenses and taxes (Elson 1990a), and men are more likely than women to spend income on luxury goods, such as alcohol and tobacco (Haddad and Hoddinott 1991; Holmboe-Ottesen and Wandel 1991; David 1991). Expenditure patterns vary from household to household. Both women and men may make necessary expenditures on welfare inputs. The key factor to note is that these expenditures tend to differ systematically, meaning that the levels of income controlled by different household members (the right hand sides of equations [5] and [6]) matters for the final mix of purchased welfare inputs (X f W and X m W ). The second social factor leading to differences in men s and women s resource allocation is that women tend to have more limited access to productive resources and markets than men. Examples of these resources are land, labor, technology, information and credit (Joekes et al 1988; World Bank 1991; Stamp 1989). Such limited access, along with constraints on women s time in other activities, constrains women s ability to engage in personal cash income generating production. 21

24 Finally, women are thought to have less bargaining power than men. This means that men s ability to influence the outcome of negotiations in their own interests is greater than women s. That Sub-Saharan African women spend more total time working than men (Koopman 1991; Mehra 1991) may be a manifestation of this balance of bargaining power in favor of men. Cultural norms and advantages and disadvantages based on gender are important factors leading to differences in women s and men s bargaining powers (Folbre 1991; Bossen 1989). In general, since women are more dependent than men on the resources which membership in households as marriage partners allows, the negative economic consequences of marital dissolution are often greater for women than for men (Blau 1989; Pollack 1985). 14 C. Allocative Inefficiency If household members had common allocational priorities, resource control and bargaining power differentials would not matter for resource allocation in households. Allocation of time would be based on these common priorities and on household members comparative advantages in different activities. Joint rationality would prevail in the maximization of household income. Income accruing to household members would be perfectly substitutable and (given expenditure responsibilities) transfers would flow freely so as to meet common allocational priorities. Income allocation would take place as if it were subject to the constraint of pooled household income. However, differences in men s and women s allocational priorities mean that resource control and bargaining power matter for resource allocation in households. They combine to produce two types of allocative inefficiency from the point of view of total household resources. The first is inefficiency in the allocation of time to income generating activities. The second is inefficiency in the allocation of time and income to welfare provisioning. Allocative inefficiency in income generation exists because of incentive incompatibilities between women and men in providing labor to communal market work (Collier 1990). Since women do not receive the product of their labor in such work, if they do not expect to gain control of income generated through transfers from men, they may have low incentive to participate. Whether women are able to resist doing so (or to 14 See Pankhurst (1991, p. 620 and footnote 5) for an example from Zimbabwe of women s post-divorce position relative to men s. 22

25 negotiate an acceptable level of income transfers) depends on many factors, among which is their bargaining power relative to men. If communal market work yields the highest returns to labor, where women are able to resist allocating labor to it in favor of their personal (less remunerative) market work, time allocation to market work is inefficient from the point of view of the household s total time endowment (Dey 1992; Jones 1983; Whitehead 1990). Both men and women provide inputs into welfare provisioning and benefit from increased levels of household welfare. However, they tend to make separate decisions--from the point of view of their own allocational priorities--regarding the allocation of the time they control to welfare provisioning (T i W ) and the income they control to welfare inputs (X i W ). In addition to this underlying factor, inefficiencies in welfare provisioning stem from 1) the non-pooling of household income; and 2) the low substitutability of women s and men s time in reproductive activities, with women s contribution generally being to the exclusion of men. Where household members have common allocational priorities, purchased inputs into welfare provisioning are chosen so as to maximize household welfare subject to the constraints of welfare provisioning technology (equation [8]) and total household income. However, if household members have differing allocational priorities and incomes are not pooled, the final mix of purchased welfare inputs may not maximize household welfare given these constraints. In this case, income allocated to welfare inputs is inefficient from the point of view of total household resources. Situations can arise where a reduction in one household member s real income (either the income they receive or the prices of goods they purchase) is not accompanied by increased transfers of cash income from other household members or changes in their expenditure patterns. A reduction in the welfare input purchases of the person whose real income is reduced could occur even when sufficient income is available in the household for these purchases. Similarly, an increase in one household member s real income may not be accompanied by increased transfers to other household members so that a corresponding increase in the latters welfare input purchases does not take place (Elson 1990b; Guyer 1988). Linked to the low substitutability of men s and women s time in reproductive activities, inefficiencies in the allocation of time to welfare provisioning may also arise. If changes in the prices associated with income generating production lead to women spending more time in market work (T f M =T f MP +T f MC increases), they may 23

26 spend less time in reproductive activities (T f R may decrease). If substitute labor is not available, this shift could have an adverse impact on household, especially children s, welfare. However, reductions in women s time spent in reproductive activities, such as child care, could be compensated by increases in income earned and spent by women on welfare inputs, such as food (Cornia, Jolly and Stewart 1987). In a review of studies in developing countries, Leslie (1988) finds that overall there is not much evidence of a negative effect of maternal remunerated employment on children s nutrition, one measure of children s welfare. However, the studies reviewed presumably refer to income that is earned and controlled by women. Women s reproductive activities are mostly to the exclusion of men, and women generally make independent decisions regarding the allocation of the time they control to such activities. However, in order to make efficient (welfare maximizing) decisions about the balance of their time between reproductive activities and market work from the point of view of their allocational priorities and constraints women need to be able to make independent decisions regarding their time in market work as well. Women do not make independent decisions over the allocation of their time to communal market work and over the allocation of the income they earn in it. Given women s lower bargaining power than men s, if women and men have differing allocational priorities, any increased time women spend in communal market work may not be accompanied by increases in the income they control. Transfers of income from men to women can be fluctuating and discretional (see Guyer 1988), nonexistent, or lower than the value of the product women produce. Thus reductions in (women s) time allocated to reproductive activities in households due to increased time in communal market work may not be compensated by increases in purchases of welfare inputs. IV. STRUCTURAL ADJUSTMENT POLICY IMPACTS ON HOUSEHOLD WELFARE In this section, using the conceptual framework presented above, I examine the possible effects of structural adjustment policies--expenditure switching policies and reductions in government expenditures on social services--on rural Sub-Saharan African households welfare. These measures can eventually impact household welfare through changing the level of purchased welfare inputs available in households and/or the amount of time spent by household members in welfare provisioning. 24

27 A. Expenditure Switching Policies Expenditure switching policies, including devaluation and liberalization, shift relative prices in favor of tradeable products. Higher prices for tradeable products give incentives to producers to reallocate productive resources, including labor, towards tradeable production. In response, agricultural households which produce tradeable crops at the onset of adjustment are expected to increase their production and sale of these crops. Households that produce nontradeable crops are expected to shift crop composition towards tradeable crops. 15 Increases in tradeables production in the agricultural sector could take place in the form of increases in the production of export crops and/or of tradeable food crops for sale on the market, both of which are cash crops. 16 Expenditure switching policies thus promote not only increased production of tradeable products, but also increased production of cash crops (as opposed to products destined for consumption in the households in which they are produced) in general. If households increase their production of cash crops, a larger proportion of their product will be in the form of cash income, and a larger proportion of their consumption, including consumption of welfare inputs, will need to be met through purchases on the market. Such tradeable-led commercialization may thus bring with it conflicts between household members surrounding income allocation, along with corresponding allocational inefficiencies in both income generation and welfare provisioning. In most rural areas of Sub-Saharan Africa, both men s and women s time are highly concentrated in agricultural production. Increases in both women s and men s time in market work may take place as a result of expenditure switching policy-induced incentives. Although the policies are pursued through changing market variables, i.e., prices, they may also have impacts on non-market variables through changing the allocation of time in households. Increases in the market work time of household members induce changes in their allocation 15 The prices of intermediate inputs into production will affect the rates of return that can be obtained. Here it will be assumed that the net effect of relative price changes is to increase the profitability of tradeable relative to nontradeable production during structural adjustment. 16 Whether food crops are tradeables or nontradeables depends on the degree to which world market prices influence their domestic prices. In some areas food is a tradeable product while in others it is not. Some staple foods produced in Sub-Saharan African countries, yams for example, are intrinsically nontradeables. Some foods that could be traded on world markets, such as many grains, are nontradeables within the rural economy due to storage and transportation costs. In countries where food produced is an unprotected import substitute, food is a tradeable commodity (Collier 1990). Food crops that are consumed primarily within households without entering the market are nontradeables. 25